What is SpiderGate?
SpiderGate is an OpenAI-compatible LLM gateway. You point your existing OpenAI client at one base URL, send a request with a task name like spideriq/coding, and SpiderGate routes it to the best available model across dozens of providers — handling auth, failover, cost tracking, and observability for you.
It is built for AI agents and the apps that run them. Anything that speaks the OpenAI chat-completions protocol works unchanged: swap the base URL and the API key, keep your code.
curl -X POST "https://spideriq.ai/api/gate/v1/chat/completions" \
-H "Authorization: Bearer $SPIDERIQ_TOKEN" \
-H "Content-Type: application/json" \
-d '{
"model": "spideriq/coding",
"messages": [{"role": "user", "content": "Write a Python function that reverses a string"}]
}'What you get
One endpoint, every provider. A single OpenAI-compatible API in front of Groq, Mistral, OpenRouter, Cerebras, NVIDIA NIM, and more — plus subscription primaries like Codex.
Task-based routing. Ask for a job (
spideriq/extraction,agent/chat) instead of a specific model. SpiderGate picks the right model and falls back automatically when one is busy or down.A key vault. Store every provider key in one encrypted place, share keys across a pool, or bring your own — no more keys scattered across
.envfiles.Per-agent control. Mint tokens with monthly budgets, rate limits, and model restrictions, and watch spend per agent.
Full observability. Every request is traced — what was sent, what came back, which model served it, latency, tokens, and cost.
Why a gateway
Calling LLM providers directly means managing a different SDK, key, rate limit, and failure mode for each one — and rewriting code every time you switch models. SpiderGate collapses that into a single, stable interface:
You write to one protocol. The OpenAI chat-completions shape, which almost every tool already speaks.
Routing is a config concern, not a code concern. Change which model serves
spideriq/codingwithout touching your app.Failover is automatic. When a provider rate-limits or errors, the request falls through a chain of alternatives before it ever fails.
When to use SpiderGate
SpiderGate is a strong fit when you want to:
Run AI agents or workers that call LLMs and need reliable, multi-provider access behind one URL.
Maximize free tiers by rotating across many provider keys before paying for premium models.
Control and observe spend per agent, with budgets, rate limits, and per-request traces.
Centralize provider keys in an encrypted vault instead of spreading them across services.
It may not be the right fit if you need a self-hosted, on-premises model-serving stack (SpiderGate is a managed gateway), or if you only ever call a single model from a single provider and don't need routing, pooling, or observability.
How this manual is organized
Getting Started — what SpiderGate is, a quickstart, and the core concepts.
Using the Gateway — chat completions, streaming, task aliases, models, and multi-modal endpoints.
Dashboard — the overview, agent keys, the vault, and the playground.
Observability — request traces.
Reference — the full API, authentication, errors, and rate limits.
Next steps
Send your first request with the Quickstart.
Learn the Core Concepts — aliases, providers, and tokens.
Browse the Task Aliases to pick the right one for your job.